A New Variable Step-Size NLMS Adaptive Filtering Algorithm

Abstract

There is a contradiction in classical adaptive filtering algorithm that fast convergence speed comparing with low steady state error. In order to improve this contradiction, this paper presents a new non-parametric Variable Step-Size NLMS algorithm. The new algorithm uses gradient vector's features to achieve the step iteration and a certain approximation process was made for it. Software simulation and hardware implementation verify the effectiveness of the algorithm.